MPLS and GMPLS are technologies for traffic engineering and provisioning. MPLS brings layer 2 switching speeds to layer 3 and supports multiple layer 2 technologies. GMPLS generalizes MPLS to support different network types including packet, TDM, and wavelength services. While MPLS may eventually replace ATM, the migration will likely be gradual due to existing network investments and immaturity of the technologies. The future success of GMPLS depends on proving its benefits for provisioning IP over optical networks.

1. MPLS and GMPLS
Li Yin
CS294 presentation

2. Outline
 Part I: MPLS
 Part II: GMPLS
 Part III: The reality check

3. Part I: MPLS

4. Why MPLS?
 MPLS stands for: “Multi-Protocol Label Switching”
 Goals:
– Bring the speed of layer 2 switching to layer 3
• May no longer perceived as the main benefit: Layer 3 switches
– Resolve the problems of IP over ATM, in particular:
• Complexity of control and management
• Scalability issues
– Support multiple layer 2 technologies

5. Basic Idea
 MPLS is a hybrid model adopted by IETF to incorporate best
properties in both packet routing & circuit switching
Forwarding:
Label Swapping
Control:
IP Router
Software
Control:
IP Router
Software
Forwarding:
Longest-match
Lookup
Control:
ATM Forum
Software
Forwarding:
Label Swapping
IP Router MPLS ATM Switch

6. Basic Idea (Cont.)
 Packets are switched, not routed, based on labels
 Labels are filled in the packet header
 Basic operation:
– Ingress LER (Label Edge Router) pushes a label in front of the IP
header
– LSR (Label Switch Router) does label swapping
– Egress LER removes the label
 The key : establish the forwarding table
– Link state routing protocols
• Exchange network topology information for path selection
• OSPF-TE, IS-IS-TE
– Signaling/Label distribution protocols:
• Set up LSPs (Label Switched Path)
• LDP, RSVP-TE, CR-LDP

7. MPLS Operation
1a. Routing protocols (e.g. OSPF-TE, IS-IS-TE)
exchange reachability to destination networks
1b. Label Distribution Protocol (LDP)
establishes label mappings to destination
network
2. Ingress LER receives packet
and “label”s packets
IP
3. LSR forwards
packets using label
swapping
4. LER at egress
removes label and
delivers packet
IP

8. Main features
 Label swapping:
– Bring the speed of layer 2 switching to layer 3
 Separation of forwarding plane and control plane
 Forwarding hierarchy via Label stacking
– Increase the scalability
 Constraint-based routing
– Traffic Engineering
– Fast reroute
 Facilitate the virtual private networks (VPNs)
 Provide class of service
– Provides an opportunity for mapping DiffServ fields onto an MPLS
label
 Facilitate the elimination of multiple layers

9. Part II: GMPLS

10. Outline
 Why GMPLS?
 GMPLS and MPLS
 Control interfaces
 Challenges of GMPLS
 Several proposed techniques
– Suggested label
– Bi-direction LSP setup
– LMP
 Summary

11. GMPLS
 GMPLS stands for “Generalized Multi-Protocol Label
Switching”
 A previous version is “Multi-Protocol Lambda
Switching”
 Developed from MPLS
 A suite of protocols that provides common control to
packet, TDM, and wavelength services.
 Currently, in development by the IETF

12. Why GMPLS?
 GMPLS is proposed as the signaling protocol for optical
networks
 What service providers want?
• Carry a large volume of traffic in a cost-effective way
• Turns out to be a challenge within current data network architecture
• Problems:
– Complexity in management of multiple layers
– Inefficient bandwidth usage
– Not scalable
• Solutions: eliminate middle layers IP/WDM
 Need a protocol to perform functions of middle layers
IP
ATM
SONET/SDH
DWDM
Carry applications and services
Traffic Engineering
Transport/Protection
Capacity

13. Why GMPLS? (Cont.)
 Optical Architectures
 A control protocol support both overlay model and peer model
will bring big flexibility
– The selection of architecture can be based on business decision
Peer Model
Overlay Model
UNI UNI

14. Why GMPLS? (Cont.)
 What we need? A common control plane
– Support multiple types of traffic (ATM, IP, SONET and etc.)
– Support both peer and overlay models
– Support multi-vendors
– Perform fast provisioning
 Why MPLS is selected?
– Provisioning and traffic engineering capability

15. GMPLS and MPLS
 GMPLS is deployed from MPLS
– Apply MPLS control plane techniques to optical switches and
IP routing algorithms to manage lightpaths in an optical
network
 GMPLS made some modifications on MPLS
– Separation of signaling and data channel
– Support more types of control interface
– Other enhancement

16. Control interfaces
 Extend the MPLS to support more interfaces other than packet
switch
– Packet Switch Capable (PSC)
• Router/ATM Switch/Frame Reply Switch
– Time Division Multiplexing Capable (TDMC)
• SONET/SDH ADM/Digital Crossconnects
– Lambda Switch Capable (LSC)
• All Optical ADM or Optical Crossconnects (OXC)
– Fiber-Switch Capable (FSC)
 LSPs of different interfaces can be nested inside another
FSC
LSC
LSC
TDMC
TDMC
PSC

17. Challenges
 Routing challenges
– Limited number of labels
– Very large number of links
• Link identification will be a big problem
• Scalability of the Link state protocol
• Port connection detection
 Signaling challenges
– Long label setup time
– Bi-directional LSPs setup
 Management challenges
– Failure detection
– Failure protection and restoration

18. Suggested label
 Problem: it takes time for the optical switch to program switch
– Long setup time
 Solution:
– Each LSR selects a label (Suggested Label) and signals this label to
downstream LSR, and start program its switch.
 reduce LSP setup overhead
Suggested Label = l1
Program Switch l1 X l2
Suggested Label = l2
Reserved Label = l3
Reserved Label = l4
Make sure the programming
request has completed
Request
Program Switch l1 X l2
Request
Map Label = l2
Map Label = l1
No suggested label with suggested label

19. Bi-Directional LSP setup
 Problem: How to set up bi-directional LSP?
 Solution:
– Set up 2 uni-directional LSP
• Signaling overhead
• End points coordination
– One single message exchange for one bi-directional LSP
• Upstream Label.
Suggested Label = l1
Upstream Label = la
Suggested Label = l2
Upstream Label = lb
Reserved Label = l3
Reserved Label = l4
la lb
l3
l4

20. Link Management Protocol
 Problem:
– How to localize the precise location of a fault?
– How to validate the connectivity between adjacent nodes?
 Solution: link management protocol
– Control Channel Management
– Link Connectivity Verification
– Link Property Correlation
– Fault Management
– Authentication

21. GMPLS Summary
 Provides a new way of managing network resources
and provisioning
 Provide a common control plane for multiple layers
and multi-vendors
 Fast and automatic service provisioning
 Greater service intelligence and efficiency

22. Part III: The Reality Check

23. Question:
Will MPLS replace ATM?

24. Opinion 1:
 MPLS might replace ATM eventually however, the
migration may be slow.
 Why MPLS will replace ATM eventually?
– Future network is data-centric
• IP instead of ATM
– MPLS can act ATM’s functionalities
• Traffic engineering using MPLS
• VPNs based on MPLS
– From service provider’s view, MPLS reduces the cost and
provides operational efficiencies
– Scalable

25. Opinion 1 (Cont.)
 MPLS deployment status
 ISPs deploy/plan to deploy MPLS for traffic engineering and VPNs
– UUNET, AT&T, Equant, Global Crossing, Cable & Wireless and etc..
 Equipment vendors are pushing MPLS to the market
 Lucent killed its next-generation ATM core switch and switch to MPLS-based
switch

26. Opinion 1 (Cont.)
 Why the migration may be slow?
– ATM is still the biggest revenue generator
• The networks are installed already
• Customers care about the price and the services only
– MPLS is more expensive
– ATM can provide most service MPLS can provide
• ISPs care more about revenue than new technologies
– ISPs have to grow their existing business. At this point, they are more
concerned about leveraging existing services rather than migrating to new
technologies for technology’s sake.
– The cost of migration
– MPLS still has problems to be solved
• Interoperability
– It takes time for a protocol to be mature. (usually 5 years)

27. Opinion 2
 MPLS cannot COMPLETELY replace ATM
 Why?
– Some customers may still choose ATM instead of MPLS
• Traffic engineering of ATM
– ATM provides better QoS than MPLS
– For those customers care about delay and jitter, they may want to
stick to ATM instead of trying a new technology
• ATM based VPN
– Customers maintain the routing table
– MPLS based VPN: entail ISP handling all the routing on behalf of
customers
– Will customer trust ISP?
– The size of the routing table.

28. GMPLS Questions
 Does the success of GMPLS depend on the success of MPLS?
– No.
– MPLS and GMPLS are proposed for different purposes.
– GMPLS is proposed to support IP over WDM. After all, a signaling
protocol is needed to perform provisioning.
 The future of GMPLS is unclear
– GMPLS certainly will offer operational benefits to carriers
• However, it is not necessarily provide immediate return on investment.
• Need to prove the efficacy
– GMPLS proposes an entirely new way of managing network
resources and provisioning
• More difficult to be adopted
 It may take some time to prove GMPLS.

29. Summary
 MPLS and GMPLS are promising technologies
 ISPs are interested in MPLS and GMPLS
 Whether the MPLS will replace ATM or not has no
final answer
 The efficacy of GMPLS may take years to prove